Family physicians might react to a new article about diabetes—such as the one by Gavin and colleagues¹ in this issue of American Family Physician—by thinking, “Another diabetes article? I know how to care for diabetes. I'll skip this one.” Please don't skip this article. Diabetes care is not only about knowledge. If it were, it would be hard to explain why 74 percent of persons with diabetes have uncontrolled blood pressure, 71 percent have elevated lipid levels, and 54 percent have hemoglobin A1c levels greater than 7 percent.²

Who is responsible for the inadequacy of diabetes management? Should physicians be blamed? Should patients? Indeed, some physicians are unaware of accepted diabetes guidelines, and patients may be resistant to changing their behavior. However, the main difficulty lies not with physicians or patients but with the health care system. Acute complaints crowd out chronic care management; the painful knee takes priority while less urgent diabetes care gets short shrift (the “tyranny of the urgent”).³

Physicians' attempts to help patients change their behavior often are not performed productively.⁴ At times, in the pressured atmosphere of primary care, routine tasks such as ordering blood tests are neglected instead of being delegated to nonphysician personnel who may have more time. Clinical information systems are rarely available, and many physicians are unable to produce a list of their patients with diabetes.

The Chronic Care Model was developed to improve the management of chronic illness.^5,6 Achieving the goals discussed by Gavin and colleagues¹ will require primary care practices throughout the country to implement this model. Some of its components include clinical information systems (e.g., registries, reminders, physician feedback), practice redesign (e.g., team care, planned visits, case management), decision support (e.g., practice guidelines, physician education), and self-management support (e.g., patient education, training patients in goalsetting skills). In diabetes management, the most important components may be the registry, reminders, planned visits, physician feedback, and self-management training.⁶

The foundation of the Chronic Care Model is the registry, which lists all of the patients on a physician's panel who have a chronic condition. A diabetes registry can be derived from practice-management software, pulling the ICD-9 codes of all patients with diabetes. In the ideal situation, data about A_1c levels and low-density lipoprotein (LDL) cholesterol levels can be put into the registry electronically, and blood pressure data can be loaded into the registry from an electronic medical record. However, because most family physicians do not have those capabilities, a medical assistant or billing clerk can input the clinical data (A_1c level, LDL cholesterol level, and blood pressure) from a diabetes flow sheet. Ideally, the registry also would track when the last eye examination and microalbumin test were performed. The registry can be used to generate reminders, provide physician feedback, and classify patients with well-controlled diabetes or poorly controlled diabetes.

Before each patient visit, a medical assistant can print a reminder prompt from the registry and (using a physician-written protocol) order laboratory tests or eye examinations that are overdue, saving the physician time and ensuring that these routine tasks are performed. Reminders are known to be effective; 22 of 26 studies on physician reminders showed improvement in physician performance.⁷

Registries also can be used to generate letters to patients who are overdue for office follow-up, laboratory tests, or eye examinations. A controlled study of registries with letters to patients found greater reductions in A_1c and LDL cholesterol levels in patients who received letters than in control patients.⁸

The registry can be used to measure over time the percentage of patients with diabetes who have A_1c levels over 8 percent, LDL cholesterol levels over 130 mg per dL (3.4 mmol per L), and blood pressure levels greater than 130/80 mm Hg. A Cochrane review⁹ has shown that this kind of physician feedback improves practice, although the effect is less than that achieved by physician reminders.

Patients in the registry can be stratified byA_1c, LDL cholesterol, and blood pressure levels, and patients with poor control can be targeted for planned diabetes visits. These visits, which circumvent the “tyranny of the urgent,” have been shown in randomized trials to reduce A_1c levels compared with control subjects.^10–12 Ideally, the visits are conducted by nurses using physician-generated protocols and combine patient education with medication management.

Finally, patient self-management training is critical to successful care of diabetes and associated hyperlipidemia and hypertension. A comprehensive review of traditional patient education found that patient knowledge increased, but glycemic control did not.¹³ In one randomized trial,¹⁴ training in goal setting and problem solving were added to traditional patient education, resulting in improved A_1c levels in the intervention group. Rather than telling patients how to live their lives, it is more productive for physicians to work collaboratively with patients, eliciting their readiness to make behavior changes and agreeing on focused behavior-change action plans.¹⁵